Research map / Image generation / Image restoration
GSNR: Graph Smooth Null-Space Representation for Inverse Problems
Romario Gualdrón-Hurtado, Roman Jacome, Rafael S. Suárez, Henry Arguello
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add GSNR null-space regularization to PnP | PGD-PnP · CelebA super-resolution x4 | PSNR (dB): 27.37 -> 29.42 | Improved |
| add GSNR null-space regularization to PnP | PGD-PnP · CelebA demosaicing | PSNR (dB): 39.35 -> 39.89 | Improved |
| replace learned arbitrary null-space basis (NPN) with graph-smooth basis L_8nn | GSNR-PnP · CelebA super-resolution x4 | PSNR (dB): 29.21 -> 29.38 | Improved |
| use 4-neighbor Laplacian instead of 8-neighbor (ablation) | GSNR-PnP · CelebA super-resolution x4 | PSNR (dB): 29.38 -> 29.42 | No clear effect |
| replace NPN with GSNR L_8nn | PGD-PnP Lip-DnCNN · CelebA deblurring | PSNR (dB): 37.86 -> 38.18 | Improved |
| replace NPN with GSNR L_8nn | PGD-PnP Wavelet · Places365 deblurring | PSNR (dB): 33.17 -> 33.59 | Improved |
| add NPN to baseline PGD-PnP | PGD-PnP Lip-DnCNN · Places365 deblurring | PSNR (dB): 31.58 -> 33.22 | Improved |
| add null-only graph penalty gamma_g (ablation) | GSNR L_8nn · CelebA deblurring | PSNR (dB): 38.18 -> 38.18 | No clear effect |
| add null-only graph penalty gamma_g=0.1 (ablation) | GSNR L_8nn PnP-PGD · Deblurring convergence | convergence speed | Improved |
| add GSNR term to DiffPIR/DPS diffusion solvers | DiffPIR, DPS · CelebA super-resolution | PSNR (dB): 30.19 -> 30.31 | Improved |
Closest papers by meaning
- NPN: Non-Linear Projections of the Null-Space for Imaging Inverse Problems (NeurIPS 2025 · similarity 0.91)
- Deep Decomposition Learning for Inverse Imaging Problems (ECCV 2020 · similarity 0.85)
- A convex program for bilinear inversion of sparse vectors (NeurIPS 2018 · similarity 0.85)
- It Has Potential: Gradient-Driven Denoisers for Convergent Solutions to Inverse Problems (NeurIPS 2021 · similarity 0.85)
- FiRe: Fixed-points of Restoration Priors for Solving Inverse Problems (CVPR 2025 · similarity 0.84)
- Provably Robust Score-Based Diffusion Posterior Sampling for Plug-and-Play Image Reconstruction (NeurIPS 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
